Why System Center Works for the Non-Enterprise: Total Visibility, Total Control

Greg Shields


At a Glance:

  • Proactive monitoring for non-enterprises
  • From monitoring to acting
  • Automated virtual administration
  • Enterprise backup features for the non-enterprise


Contents
Proactive Monitoring for Non-Enterprises
From Monitoring to Acting
Automated Virtual Administration
Enterprise Backup Features for the Non-Enterprise
Visibility and Control that Fits



Travis Morrison, senior systems administrator with New Belgium Brewery, a Fort Collins, Colo., specialty brewer most known for its Fat Tire Ale, knows first-hand the power of total network visibility. As a well-known business that's not small but definitely not "enterprise," New Belgium needs constant and assured access to its IT infrastructure. Its servers and workstations enable the company to work with customers, ensure product deliveries and manage a complex manufacturing schedule.
Not long ago, Morrison found himself dealing with recurring problems in one of New Belgium's sales applications, a complex solution that integrates SQL Server, IIS and internal batch processing.
When the sales application encountered a problem, it stopped functioning, often requiring IT to spend a full day restoring its services. . New Belgium employees were unable to work with customers. The sales team couldn't view sales records or . locate data necessary for alerting customers of trending changes. Clearly, this wasn't good for business. After a painful and expensive series of outages, Morrison decided he needed a proactive approach that would allow him to root out problems before users noticed them. After researching multiple products, he settled on System Center Operations Manager 2007 R2. "The system quite literally paid for itself in less than 24 hours. With a little help, we were able to get the server up and running, agents installed throughout the network and alerting to our pagers turned on in less than a day," Morrison says.
"That very night, we were paged at around 3 a.m. to find out that the sales application database was running out of space. It turns out that this was the problem all along. The difference this time is that we were able to shut down SQL Server and give it the additional space it needed without causing yet another all-day outage at cost to the brewery," says Morrison in describing New Belgium's immediate return on investment for Operations Manager.

Proactive Monitoring for Non-Enterprises
Morrison's story is particularly interesting because. New Belgium isn't your typical System Center enterprise customer. Although salespeople do travel throughout the United States, New Belgium runs only one brewery in a single location. As such, it's not what you'd consider the classic enterprise IT organization. Rather, it's more of a small or midsize business, with small and midsize needs. Nonetheless, New Belgium's need for a highly available IT infrastructure is much like that of the typical enterprise. It needs as close to 100 percent uptime as it can get for core services like e-mail, file servers and customer databases. It needs to know when problems occur, preferably before users call in to complain. It needs a comprehensive monitoring solution that works across its heterogeneous (though mostly Microsoft-centric) IT environment.
Operations Manager enables just that kind of solution in a package that works for Morrison's environment. Operations Manager creates a platform for monitoring system behaviors across servers and workstations, as well as other endpoints like storage and networking. An Operations Manager infrastructure is designed to scale to meet the needs of any size network. And what's best is that Operations Manager is but one part of Microsoft's System Center portfolio. Other pieces include System Center Configuration Manager for configuration management, System Center Virtual Machine Manager for unified virtual machine administration and System Center Data Protection Manager for enterprise-quality backup and restore.
For the 350-user, 40-server size of Morrison's environment, installing Operations Manager on a single server meets performance expectations. An instance of SQL Server is installed on the same or another server in the environment for the storage of Operations Manager monitoring data. Operations Manager is quite unlike most other monitoring solutions, which enable a central location for consolidating and analyzing Event Log and system performance data. Operations Manager takes this process a step further through its use of Management Packs (MPs). Designed by Microsoft as well as other product vendors for their own hardware and software, these MPs are essentially sets of rules that filter through the tidal wave of incoming raw data (see Figure 1) to alert administrators of only those behaviors likely to interest them.
Figure 1 Management Packs filter through raw data to show actionable information in the Operations Manager console. (Click the image for a larger view)

Think about the types of behaviors your IT organization needs to know about as soon as possible. Is a server running out of disk storage? Does an Event Log entry contain information about a failing application? Has performance on a key server dropped below acceptable thresholds? Finding out about these behaviors as they occur can help your organization address problems before they impact the business.
However, today's IT infrastructures are ridiculously complicated, and this is especially problematic for non-enterprise organizations, which tend to have smaller IT teams. It means that individual IT professionals must be proficient in multiple technologies across the infrastructure rather than being able to specialize in just a few. Tracking down the root causes of every problem is nearly impossible for any IT individual, even as such is expected.
This is where the MPs can really help. MPs, which are typically created by a product's development team or by consultants who have hundreds of engagements under their belts, filter through raw data to illuminate just the behaviors important to administrators. Using overrides and thresholds, an IT organization can customize an MP to deliver the kinds of information it needs most. For example, if you're not terribly concerned about processor use going over 80 percent but want to know immediately when it hits 95 percent, you can customize a threshold on the Operations Manager monitor to reflect that sensitivity. You can adjust specific alerts for server behaviors by setting an override for that alert's characteristics and applying the override to individual servers or the entire environment.
IT professionals can even use Operations Manager for monitoring non-Microsoft components, such as the network itself. At Bynet Data Communications Ltd., Operations Manager's network integration helped quantitatively refute user complaints that the network was slow, says Idan Yona, management solutions specialist at the company. "Our Operations Manager installation saved the day in its initial deployment. We deployed Operations Manager with a particular target in mind, which was to estimate the users' network experience in our remote offices. After the system was up for a short period of time, we were able to gather network statistics that compared the users' experience between the main site and our branch offices. The resulting data disproved our remote users' claims that they were experiencing network slowness, enabling us to cancel a major WAN upgrade project. The result was a savings to the company of more than $60,000," he says.
The Microsoft MP catalog, which includes detailed information about the types of monitoring enabled by each MP (see Figure 2), is available online. The number and types of MPs are extensive, and even the most heterogeneous of organizations will find the right kinds of integrations for their needs, even into third-party technologies not commonly associated with Microsoft. Many MPs are downloadable for free, while others are available for purchase from their particular vendor. Each brings another set of data into an Operations Manager infrastructure for proactively alerting when issues occur.
Figure 2 The System Center Management Pack catalog provides detailed listings. (Click the image for a larger view)


From Monitoring to Acting
Monitoring an environment gets you only partway to a goal of total control. A second critical capability is the ability to enact change on servers and workstations in a trackable, repeatable and predictable fashion. Organizations that find themselves deploying technicians to user desks for software installations, common troubleshooting or patch management issues cannot be successful in managing a small network as it scales. This takes too many manual actions that cannot be replicated across a large swath of machines. What's needed are centralized tools for deploying software, updates and even complete operating systems to any number of computers at the same time.
For Eric Schmidt, senior systems engineer at a midsize defense contractor in Colorado, System Center Configuration Manager brought security to an otherwise-insecure environment of nearly 2,000 desktops, he says.
"Before [Configuration Manager,] we tried a number of different patching solutions, with varying success. Some solutions worked great for Microsoft patches, but couldn't deploy updates for our non-Microsoft software. Our environment has multiple owners and multiple IT organizations, all on the same network. So, entire sets of solutions couldn't fit into our organizational model. The least useful were those that involved a lot of manual effort with zero capabilities for unified reporting. At our worst, we guessed that we were maybe 20 percent compliant to our needed patches. Something had to change." Configuration Manager solved Schmidt's problem by enabling an automated update management infrastructure that worked with the company's multiple IT teams. "With [Configuration Manager,] we were able to create multiple collections of computers, each managed by a designated set of IT administrators," he says. "Because [Configuration Manager] natively integrates with Windows Server Update Services, each IT team could easily identify which patches made sense for their computers. When we found highly critical updates that needed immediate deployment to all computers, we would simply push them to our All Computers collection."
To do this, Configuration Manager creates a dynamic grouping of computers through what it calls collections. Unlike other solutions where administrators create groups by computer name or other static metrics, Configuration Manager's collections are based on each computer's dynamic characteristics. It is just as easy to create a grouping of computers according to where they're located on an IP subnet as it is to create one for "all computers with Microsoft Office Word 2007 installed" (see Figure 3). As a computer's characteristics change over time, an administrator may regularly add and remove it from collections. This fluidity is a major benefit of Configuration Manager. Collections are tiered, so an administrator can create subcollections that leverage additional constraints. In Schmidt's case, top-level collections relate to the computer's IT team owner, with each team granted permissions to create subcollections beneath.
Figure 3 Confi guration Manager lets you create a dynamic collection that, for example, groups all computers with Microsoft Office Word 2007 installed. (Click the image for a larger view)

While Configuration Manager is great for deploying Microsoft updates, its true power is its ability to handle software or configuration changes for non-Microsoft products as well. Schmidt took advantage of such capability in his environment. "With [Configuration Manager,] we could deploy updates to Rational ClearCase or Adobe Acrobat using the same infrastructure as we did for patches. Once an administrator got the hang of creating silent 'packages' for automated distribution, one person could deploy essentially any change to any number of computers all at once," he says.
Brita Rood, IT systems analyst for the city of Bellevue, Wash., also turned to Configuration Manager's deployment capabilities for non-Microsoft products. "We use Configuration Manager's software distribution feature to push third-party software products to computers. One example was a product that could take up to an hour to install manually. We were able to automate the installation and push it out to over 70 clients overnight. This saved time and a substantial amount of money because we didn't have to pay an outside service vendor to assist with the installations," she says.
The Configuration Manager change-management solution is scalable, so it can be used in small environments or in large enterprises serving tens of thousands of users and hundreds of sites. No matter how large the environment, the same processes and actions work in deploying changes. Need to deploy a piece of software or an entire operating system to a set of computers? Simply create your installation package using native or third-party tools, then ingest it into the Configuration Manager infrastructure. Using the console, a deployment package will combine with an advertisement and a collection to create a change event. Here, the collection defines which computers are targeted for the package, with the advertisement identifying the schedule for deployment. Using advertisements and their associated maintenance windows in Configuration Manager, an administrator can schedule deployments to occur only during periods of inactivity, such as after hours.
Once software is deployed, successfully managing it requires knowing where it is installed as much as troubleshooting it when it fails. Knowing the hardware and software inventory of a midsize IT environment is something often done through pen and paper and trial and error in many reactive-mode environments. Using Configuration Manager's inventory functions in combination with its Software Metering and Asset Intelligence features almost completely automates the procedure.
Many tools today can interrogate servers and workstations to identify their hardware and software compositions. While useful, this information is only a static representation. Configuration Manager's Software Metering feature adds the capability to identify which users and computers are actually using installed software. With Software Metering, an IT organization can locate unused software licenses, reclaiming them for use by others instead of buying additional licenses. Integrating this capability with Configuration Manager's built-in Asset Intelligence database, IT can match the individual characteristics of installed software to specific products, version numbers and editions. The result is a workable reporting engine for identifying exactly what kinds of software are installed on any managed computer.
Using these capabilities, the city of Bellevue quickly "trued up" its licensing while eliminating wasteful spending on unused licenses, Rood says. "The reporting through Asset Intelligence along with Software Metering will help us determine our specific software usage, especially for our third-party software to determine our licensing needs."
Many organizations also use Configuration Manager to deploy an entire OS to managed workstations. This complex and often expensive activity can be a major roadblock for many organizations that want to migrate to a modern Microsoft OS. Reasons for the complexity lie within the deployment itself. Traditional upgrades require technicians to handle workstations individually, often through the use of imaging software or script-based solutions. Either way, user data from the old system installation isn't preserved to the next. Thus, technicians must work with users to ensure that their specific settings (a workstation's "personality," if you will) migrate fully and correctly.
Configuration Manager's Operating System Deployment (OSD) feature, built into the base product, provides a comprehensive approach to the managed deployment of an OS. Using OSD, an administrator can create a generalized installation of the Windows operating system, which can be installed to virtually any desktop, laptop or server. Using the built-in Plug-and-Play components natively available on every modern Windows OS, device drivers can be specifically targeted for each class of hardware.
Device drivers are only one part of OS deployment. As noted earlier, maintaining a user's personality information is also critically important when upgrading from one OS to another or refreshing a broken workstation. Configuration Manager's OSD feature includes the capability to offload user settings automatically before a computer is refreshed. Once the OS installation is complete, those user settings are then reapplied. The end result looks remarkably similar to the "old" OS instance from the user's perspective, allowing a quick return to work.
Cleber Marques, technical consultant with 2S Inovações Tecnológicas in São Paulo says he has used this level of automation many times. "Using Configuration Manager, our customers can distribute Windows Vista to all of their 1,000 or more desktops, and fully expect to be in compliance with a main office in just a few days," he says.

Automated Virtual Administration
While Operations Manager and Configuration Manager help on the physical server front, many IT organizations today also are grappling with virtual servers. Virtualization seems to be everywhere these days, with environments of all sizes moving from physical to virtual machines in droves. Early adopters of virtualization were limited in the options available for virtual platforms, but today's explosion in interest has generated an equivalent increase in the options available. You can argue that the virtual wars between platform vendors are quickly approaching a draw when it comes to the feature sets available in the hypervisor itself. Today's battlefront really deals more with the tools available that help you manage your virtual infrastructure.
To that end, Microsoft has elected to take a multiplatform approach with its virtual management solution, System Center Virtual Machine Manager (VMM). Using VMM, an environment isn't limited to managing virtual machines atop just a single hypervisor. IT administrators can manage virtual infrastructures comprising the Microsoft Hyper-V as well as VMware's ESX and vCenter products from within the same console screen.
This means that environments that made an early move to virtualization with VMware products can easily manage a Hyper-V infrastructure as well, all inside the same VMM management console. Figure 4 shows a view of how that console might look when Hyper-V and ESX hosts are connected. As you can see, the display of Host Groups doesn't discriminate between the two different hypervisors. Within VMM, an administrator can invoke actions irrespective of the underlying virtual platform.
Figure 4 A view of VMM Host Groups that include Hyper-V and ESX servers. (Click the image for a larger view)

Deploying virtual machines (VM) in a Microsoft environment. requires installation of the Hyper-V hypervisor, found natively within the Window Server 2008 operating system. Once the Hyper-V role is installed to a Windows Server 2008 instance, that server can host as many VMs as its resources will allow.
The native Windows Failover Clustering adds high availability to the mix. This feature is a separate Hyper-V component for virtual servers that will participate in a cluster. Clustering enables VMs from multiple Hyper-V hosts to fail over to alternate hosts when problems occur. . If you haven't worked with Windows Failover Clustering in a few OS versions, now is the time to reintroduce yourself. Installation and management improvements make clustering a set-it-and-forget-it part of your environment.
Neither Windows Failover Clustering nor VMM are required to manage Hyper-V VMs. Their presence just makes management dramatically easier, especially as your count of Hyper-V hosts increases. With VMM, you can manage Hyper-V hosts and their VMs as units, performing actions across one or all at the same time. VMM, in effect, enables the same kinds of automation for your virtual environment that Configuration Manager and Operations Manager do for your individual server instances.
Tim Clauson, IT manager for the city of Roseburg, Ore., says he came to appreciate VMM's flexibility early on, . after a challenging and ultimately unsuccessful attempt at rolling out a different solution. "After two weeks [on site,] two consultants from [another virtual platform vendor] still hadn't gotten a single virtual machine to power on. We finally asked them to go home and moved our focus to Hyper-V and VMM. In literally six hours, we were up and running with our first virtual machines. As one of only two IT personnel for the city, I am forced to be a technology generalist and master of none. VMM was trivial for us to understand, and easy to work with," he says.
VMM's built-in Physical-to-Virtual (P2V) functionality, even in its beta versions, proved particularly helpful, Clauson says. "We loved P2V with VMM because it worked flawlessly. We wanted to be an early adopter of this technology. So we used the P2V functions of VMM's beta to virtualize our production Exchange 2007 server onto Hyper-V. That first P2V was so successful that we ended up running our mail services atop Hyper-V from that moment on," he adds.
Since that initial installation, Clauson says that he has successfully virtualized the city's SQL Server and other services atop Hyper-V and a fully licensed VMM instance. And, since he did not need to use money originally earmarked for spending on the other virtualization technology, he's been able to redirect those funds to other needy projects. This is a boon given tax shortfalls and the city's declining budget, he adds.

Enterprise Backup Features for the Non-Enterprise
A fourth System Center solution that hasn't yet gotten much penetration into IT organizations is Data Protection Manager (DPM). This relatively new offering, with a critical R2 release only recently announced, provides a platform for backing up servers and desktops throughout the IT environment. DPM stands apart from other backup solutions in its integration with other Microsoft and System Center products as well as its flexibility to back up server data to disk and tape.
DPM's focus on disk-to-disk backups is particularly beneficial to non-enterprise environments where large-scale tape arrays with robot support may not be affordable or even desired. In a world where the cost of disk drives is shrinking dramatically every day, backing up critical servers to disk as opposed to tape enables a number of unique benefits, including the rapid restoration of individual files or entire servers directly from disk backups.
Using DPM's continuous data protection integrations with Microsoft and third-party applications such as Exchange Server (see Figure 5), SQL Server or SharePoint Server, DPM can protect data from corruption or deletion in almost real time. IT managers can set short- and long-term goals for data protection, enabling multiple targets for data protection depending on need. Further, since DPM is a member of the System Center family, administrators can monitor its behaviors natively through Operations Manager.
Figure 5 DPM lets you confi gure short-term goals for an Exchange server backup. (Click the image for a larger view)

"DPM is so up-to-date with the level of data it captures from, for example, your Exchange or SQL transaction logs, that you almost have to 'trick' it to make it lose data. Even better, if you do lose data, it is likely that you're probably only going to lose a miniscule amount," notes James Conrad, an Accusource consultant who regularly helps clients with DPM installations.
While disk-to-disk backups seem useful for quick restores, most organizations require some archival of data to off-site storage. Usually, this means rotating tapes to an off-site storage facility. DPM changes this workflow by enabling backups to go from disk to disk and then later to tape media.
"For performance purposes, many people use DPM to create a disk-to-disk-to-tape architecture. By doing this, you can immediately restore lost files from disk backups, because they're directly available on the backup disks," Conrad says. "This makes the hassle of finding and loading the right tapes unnecessary. If you need to store data off-site for long-term storage, DPM provides a way to automate the disk-to-tape portion, replicating data to tape for other purposes. This multiple-hop approach eliminates the performance bottlenecks usually seen in a strictly tape-oriented solution."

Visibility and Control that Fits
Providing monitoring, configuration control, virtual platform management and backup support, the four major components of System Center can indeed satisfy the needs of virtually any IT operation. While much of System Center's press relates to its use in large, enterprise environments, the product is also a fit for the medium-sized businesses that comprise a substantial chunk of industry. If you consider yourself a reactive IT organization, dealing with problems as they happen and finding out about issues when users call to complain, consider Microsoft's System Center solutions as your path to total visibility and total control.

Greg Shields, MVP, is a partner at Concentrated Technology. Get more of Greg's Jack-of-all-Trades tips and tricks at ConcentratedTech.com




موضوعات مشابه: